Laboratorio del Futuro | September 10, 2026
In July 2026, the European Union approved the Digital Omnibus Regulation on AI, the first reform of its landmark AI Act. Regulation (EU) 2026/1744 entered into force on July 27, 2026, introducing significant changes to deadlines and certain obligations. The application of requirements for high-risk systems—those that may affect safety, fundamental rights or employment—is postponed until December 2027 for standalone systems and August 2028 for systems integrated into products.
The European model is the most comprehensive in the world: a risk-based approach that imposes ex ante obligations on developers, with fines of up to 3% of global revenue for those who fail to document how their systems were trained. Among the new provisions, the ban on non-consensual sexual deepfakes and AI-generated child sexual abuse material, scheduled for December 2026, stands out.
China has followed a different path. Since September 2025, it has required all AI-generated content to be identified with visible or technical markers. In July 2026, it published an action plan on international ethical governance of AI during the World AI Conference in Shanghai. The document calls on countries to take measures ‘in light of their national conditions’ in areas such as ethical governance throughout the AI life cycle and the prevention and control of risks by category and level. China has also launched a pilot program for ethical review and AI services, with the aim of establishing a national network of services for monitoring ethical risks.
The Chinese model integrates AI governance into a state-controlled economic and security strategy, positioning the technology as a central pillar of its industrial policy.
The United States lacks a comprehensive federal AI law. The US model is market-driven, emphasizing private-sector leadership and sector-specific regulation. The legislative debate—which would require developers to incorporate safety mechanisms such as kill switches—has not reached a political consensus.
The UN warns that regulation is advancing ‘unevenly’ and that national frameworks, technical standards and bilateral agreements are not enough to contain a technology that respects no borders. The question no government answers: who has the authority—and the capacity—to stop a race that every participant fears losing?
